40 Intermediate AI Tutorials
1. Deep Learning Basics
Deep learning uses multi-layered neural networks for complex tasks.
Example: Simple deep learning model.
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense
model = Sequential([Dense(10, activation='relu', input_dim=2), Dense(1)])
print("Deep learning model created.")Deep learning model created.
Note: Deep learning excels in tasks like image and speech recognition.
2. Convolutional Neural Networks
CNNs are used for image processing tasks.
Example: Basic CNN setup.
from tensorflow.keras.layers import Conv2D
model = Sequential([Conv2D(32, (3, 3), activation='relu', input_shape=(28, 28, 1))])
print("CNN for image tasks.")CNN for image tasks.
Note: CNNs extract spatial features from images.
3. Recurrent Neural Networks
RNNs handle sequential data like time series or text.
Example: Simple RNN layer.
from tensorflow.keras.layers import SimpleRNN
model = Sequential([SimpleRNN(10, input_shape=(None, 1))])
print("RNN for sequential data.")RNN for sequential data.
Note: RNNs are ideal for time-dependent data.
4. Transfer Learning
Transfer learning reuses pre-trained models for new tasks.
Example: Using a pre-trained model.
from tensorflow.keras.applications import VGG16
base_model = VGG16(weights='imagenet', include_top=False)
print("Using VGG16 for transfer learning.")Using VGG16 for transfer learning.
Note: Transfer learning saves training time.
5. Feature Selection
Feature selection identifies the most relevant variables.
Example: Selecting features with scikit-learn.
from sklearn.feature_selection import SelectKBest
selector = SelectKBest(k=2)
print("Selecting top 2 features.")Selecting top 2 features.
Note: Reduces model complexity and overfitting.
6. Hyperparameter Tuning
Hyperparameter tuning optimizes model settings for better performance.
Example: Grid search for tuning.
from sklearn.model_selection import GridSearchCV
from sklearn.svm import SVC
param_grid = {'C': [0.1, 1, 10]}
grid = GridSearchCV(SVC(), param_grid)
print("Tuning SVM parameters.")Tuning SVM parameters.
Note: Use GridSearchCV or RandomizedSearchCV.
7. Ensemble Methods
Ensemble methods combine multiple models for better predictions.
Example: Random Forest.
from sklearn.ensemble import RandomForestClassifier
model = RandomForestClassifier(n_estimators=100)
print("Random Forest ensemble model.")Random Forest ensemble model.
Note: Boosting and bagging improve accuracy.
8. Data Augmentation
Data augmentation increases dataset size with transformations.
Example: Image augmentation.
from tensorflow.keras.preprocessing.image import ImageDataGenerator
datagen = ImageDataGenerator(rotation_range=20)
print("Augmenting images with rotation.")Augmenting images with rotation.
Note: Common for image and text data.
9. Regularization Techniques
Regularization prevents overfitting by penalizing complexity.
Example: L2 regularization.
from tensorflow.keras.regularizers import l2
model = Sequential([Dense(10, kernel_regularizer=l2(0.01))])
print("L2 regularization applied.")L2 regularization applied.
Note: Use L1 or L2 for simpler models.
10. Advanced Data Preprocessing
Advanced preprocessing handles complex data cleaning tasks.
Example: Encoding categorical data.
from sklearn.preprocessing import OneHotEncoder
encoder = OneHotEncoder(sparse_output=False)
encoded = encoder.fit_transform([['red'], ['blue']])
print(encoded)[[1. 0.] [0. 1.]]
Note: Use encoding for categorical variables.
11. Cross-Validation
Cross-validation assesses model performance robustly.
Example: K-fold cross-validation.
from sklearn.model_selection import cross_val_score
from sklearn.linear_model import LinearRegression
model = LinearRegression()
scores = cross_val_score(model, [[1], [2]], [10, 20], cv=2)
print(scores)[1. 1.]
Note: Use 5 or 10 folds for reliable results.
12. NLP Tokenization
Tokenization splits text into meaningful units.
Example: Advanced tokenization.
from nltk.tokenize import word_tokenize
import nltk
nltk.download('punkt')
tokens = word_tokenize("AI is transforming industries.")
print(tokens)['AI', 'is', 'transforming', 'industries', '.']
Note: Tokenization is the first step in NLP.
13. Word Embeddings
Word embeddings represent words as vectors for NLP.
Example: Using pre-trained embeddings.
from gensim.models import Word2Vec
model = Word2Vec([['AI', 'is', 'great']], vector_size=10, window=5, min_count=1)
print(model.wv['AI'])(Vector representation of 'AI')
Note: Use GloVe or Word2Vec for embeddings.
14. Text Classification
Text classification assigns labels to text data.
Example: Sentiment classifier.
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.linear_model import LogisticRegression
vectorizer = TfidfVectorizer()
model = LogisticRegression()
print("Text classifier setup.")Text classifier setup.
Note: Use TF-IDF for text feature extraction.
15. Image Preprocessing
Image preprocessing prepares images for AI models.
Example: Resizing images.
from tensorflow.keras.preprocessing.image import img_to_array, load_img
img = load_img('image.jpg', target_size=(224, 224))
img_array = img_to_array(img)
print("Image resized to 224x224.")Image resized to 224x224.
Note: Normalize pixel values for better results.
16. Object Detection Basics
Object detection identifies and locates objects in images.
Example: YOLO setup.
print("YOLO model for object detection.")YOLO model for object detection.
Note: Use pre-trained models like YOLO or Faster R-CNN.
17. Model Interpretability
Model interpretability explains AI model decisions.
Example: SHAP values.
import shap
explainer = shap.TreeExplainer(model)
print("SHAP explains model predictions.")SHAP explains model predictions.
Note: Use SHAP or LIME for interpretability.
18. Time Series Forecasting
Time series forecasting predicts future values based on past data.
Example: ARIMA model.
from statsmodels.tsa.arima.model import ARIMA
model = ARIMA([1, 2, 3, 4], order=(1, 0, 0))
print("ARIMA for time series.")ARIMA for time series.
Note: Use ARIMA or LSTM for forecasting.
19. Autoencoders
Autoencoders compress and reconstruct data for learning.
Example: Simple autoencoder.
from tensorflow.keras.layers import Dense
model = Sequential([Dense(10, activation='relu'), Dense(2), Dense(10)])
print("Autoencoder for data compression.")Autoencoder for data compression.
Note: Used for denoising or dimensionality reduction.
20. GANs Introduction
GANs generate new data using a generator and discriminator.
Example: Basic GAN setup.
from tensorflow.keras.models import Sequential
generator = Sequential([Dense(128, activation='relu')])
print("GAN generator model.")GAN generator model.
Note: GANs are used for image generation.
21. Advanced Model Evaluation
Advanced evaluation uses metrics like ROC-AUC for performance.
Example: ROC-AUC score.
from sklearn.metrics import roc_auc_score
score = roc_auc_score([1, 0], [0.9, 0.1])
print(score)1.0
Note: Use ROC-AUC for classification tasks.
22. Handling Imbalanced Data
Imbalanced data techniques address skewed class distributions.
Example: Oversampling with SMOTE.
from imblearn.over_sampling import SMOTE
smote = SMOTE()
print("SMOTE balances dataset.")SMOTE balances dataset.
Note: Use SMOTE or undersampling for balance.
23. Dimensionality Reduction
Dimensionality reduction simplifies data while retaining information.
Example: PCA.
from sklearn.decomposition import PCA
pca = PCA(n_components=2)
print("PCA reduces dimensions.")PCA reduces dimensions.
Note: PCA or t-SNE for visualization.
24. Advanced Clustering
Advanced clustering groups data with complex algorithms.
Example: DBSCAN clustering.
from sklearn.cluster import DBSCAN
dbscan = DBSCAN(eps=0.5, min_samples=5)
print("DBSCAN for complex clusters.")DBSCAN for complex clusters.
Note: DBSCAN handles non-spherical clusters.
25. Model Deployment with Flask
Flask deploys AI models as web services.
Example: Flask app for model.
from flask import Flask
app = Flask(__name__)
@app.route('/')
def predict():
return "Model prediction"
print("Flask app running.")Flask app running.
Note: Use Flask for lightweight deployment.
26. AI Pipeline Automation
Pipeline automation streamlines AI workflows.
Example: Pipeline with scikit-learn.
from sklearn.pipeline import Pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.svm import SVC
pipeline = Pipeline([('scaler', StandardScaler()), ('svm', SVC())])
print("Automated AI pipeline.")Automated AI pipeline.
Note: Pipelines ensure consistent preprocessing.
27. Attention Mechanisms
Attention mechanisms focus on relevant data in sequences.
Example: Attention layer.
from tensorflow.keras.layers import Attention
attention = Attention()
print("Attention layer for sequences.")Attention layer for sequences.
Note: Used in NLP and vision tasks.
28. Transformers Intro
Transformers use attention for advanced NLP tasks.
Example: Transformer model.
from transformers import BertModel
model = BertModel.from_pretrained('bert-base-uncased')
print("BERT transformer model.")BERT transformer model.
Note: Use Hugging Face for transformer models.
29. Reinforcement Learning Basics
Reinforcement learning trains agents through rewards.
Example: Q-learning setup.
import numpy as np
q_table = np.zeros((10, 2))
print("Q-learning table initialized.")Q-learning table initialized.
Note: Used in games and robotics.
30. AI Model Optimization
Model optimization improves performance and efficiency.
Example: Model pruning.
from tensorflow.keras.models import Sequential
model = Sequential([Dense(10)])
print("Model before pruning.")Model before pruning.
Note: Pruning reduces model size.
31. Handling Missing Data
Handling missing data ensures robust model training.
Example: Imputation with mean.
from sklearn.impute import SimpleImputer
imputer = SimpleImputer(strategy='mean')
data = imputer.fit_transform([[1, None], [2, 3]])
print(data)[[1. 3.] [2. 3.]]
Note: Use mean, median, or KNN imputation.
32. Sentiment Analysis
Sentiment analysis classifies text as positive or negative.
Example: Simple sentiment classifier.
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.linear_model import LogisticRegression
vectorizer = TfidfVectorizer()
model = LogisticRegression()
print("Sentiment analysis setup.")Sentiment analysis setup.
Note: Use labeled datasets for training.
33. Named Entity Recognition
NER identifies entities like names and places in text.
Example: NER with spaCy.
import spacy
nlp = spacy.load("en_core_web_sm")
doc = nlp("Apple is in California.")
print([(ent.text, ent.label_) for ent in doc.ents])[('Apple', 'ORG'), ('California', 'GPE')]
Note: Use spaCy or NLTK for NER.
34. Advanced Visualization
Advanced visualization creates insightful AI data plots.
Example: Heatmap with Seaborn.
import seaborn as sns
import numpy as np
data = np.random.rand(10, 10)
sns.heatmap(data)
print("Heatmap visualization.")Heatmap visualization.
Note: Use Seaborn for complex visualizations.
35. Model Debugging
Model debugging identifies and fixes errors in AI models.
Example: Checking model predictions.
from sklearn.linear_model import LinearRegression
model = LinearRegression()
model.fit([[1], [2]], [10, 20])
print(model.predict([[3]]))[30]
Note: Debug with validation data and metrics.
36. AI in Cloud
Cloud platforms host and scale AI models.
Example: AWS SageMaker setup.
print("AWS SageMaker for model training.")AWS SageMaker for model training.
Note: Use AWS, GCP, or Azure for cloud AI.
37. Fine-Tuning Models
Fine-tuning adapts pre-trained models to specific tasks.
Example: Fine-tuning BERT.
from transformers import BertForSequenceClassification
model = BertForSequenceClassification.from_pretrained('bert-base-uncased')
print("Fine-tuning BERT.")Fine-tuning BERT.
Note: Fine-tune with small datasets for efficiency.
38. AI Security Basics
AI security protects models from attacks like adversarial examples.
Example: Adversarial robustness.
print("Add noise to inputs to test robustness.")Add noise to inputs to test robustness.
Note: Use adversarial training for security.
39. Multi-Modal AI
Multi-modal AI combines text, images, and other data.
Example: Multi-modal input.
print("Combine text and image inputs for AI.")Combine text and image inputs for AI.
Note: Used in advanced applications like CLIP.
40. Building an AI Project
Building an AI project involves planning, coding, and deployment.
Example: Simple project pipeline.
from sklearn.pipeline import Pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.linear_model import LogisticRegression
pipeline = Pipeline([('scaler', StandardScaler()), ('clf', LogisticRegression())])
print("AI project pipeline.")AI project pipeline.
Note: Plan data, model, and deployment steps.